Abstract
OBJECTIVE
This study was performed to develop a program for predicting individual cancer risk and to validate its discrimination power between case and control groups.
METHODS
The author used the five databases for searching journals about risk factors of six major cancers in Koreans: stomach, liver, colorectal, breast, uterine cervix and lung cancer. The risk models were selected from journals presenting a multivariate linear logistic regression analysis. The baseline hazards which had no risk factors were calculated, and a cancer risk assessment program was developed using relative risks based on risk factors' combination and baseline hazards. Case-control study was performed for five years to validate the program.
RESULTS
The discrimination power between case and control was 0.827 in stomach cancer, 0.949 in liver cancer, 0.594 in colorectal cancer, 0.587 in breast cancer, 0.708 in uterine cervix cancer and 0.663 in lung cancer. The estimated cancer probabilities were higher in all case groups compared to the control groups.